Latin: Spacing
Stephen Nixon pointed out in a review that the capitals were spaced somewhat loosely.
Thin.pdf Regular.pdf ExtraBold.pdf
Jacques Le Bailly responded:
I had a quick view. And I think you are right, the caps area little bit too loose.
I made some adjustments in the characters A/H/I/M/N/O in this file. Maybe you might find it too tight, but you will see the difference. I did it for all masters, based on the spacing of the lowercase. The Black lowercase is very tight, I think. So is your space character :P
One tip if you work in Glyphs. You should use Metrics keys. Makes everything a lot easier. I added some on the characters I changed.
I asked:
The cap spacing instantly looks much better and has a similar spacing texture to the lower-case :) It's magic. I completely forgot about side-linking, which makes things easier indeed.
I take it I can apply the “I” LSB treatment to BEFKLPR and the “C” LSB to GQ, too?
Would you also suggest applying the same RSB to U and J?
Should I apply the “n” RSB to the straight side of “bdpqgklu”? The LSB to “a”?
Any useful linkings for “vwxy” and “VWXY”?
I noticed the “e” has approximately “o” side bearings in the regular, but has a tighter RSB in the light and looser one in the extra bold. Is that to be expected? At least the LSB could be linked to “o”.
Also, should the space glyph be the same in all weights generally?
Jacques Le Bailly responded:
Modification were only done on the caps. Lowercase is untouched.
The cap spacing instantly looks much better and has a similar spacing texture to the lower-case :) It's magic. I completely forgot about side-linking, which makes things easier indeed.
In my book it is all about rhythm. If you want your all caps setting to be wider, you can add a feature for that. But I would basically always link the spacing to the lowercase.
I take it I can apply the “I” LSB treatment to BEFKLPR and the “C” LSB to GQ, too?
I or H, whatever suits you best. I would take the O as base reference instead of the C.
Would you also suggest applying the same RSB to U and J?
Depends on how similar the forms are. You could also make a formula. Like RSB J is =U-5 for example.
Should I apply the “n” RSB to the straight side of “bdpqgklu”? The LSB to “a”?
For example. If you want them to be similar.
Any useful linkings for “vwxy” and “VWXY”?
Those are always difficult. For W you could use a formula, like =V+2
I noticed the “e” has approximately “o” side bearings in the regular, but has a tighter RSB in the light and looser one in the extra bold. Is that to be expected? At least the LSB could be linked to “o”.
I often link the RSB of the e to the o.
Space character should not be equal in all weights. Space in Bold is wider than in Regular etc.
I keyed “i” and “j” to “n” as well and will experiment with space some. Do you think the extra bold spacing is too tight given that it should be used for headings?
If your question is if you can mix two ways of spacing in one weight axis, middle weights for text and extreme weights for display, I wouldn’t do it. Choose one.
Depends.
Is it for display or text ?
What is your opinion/taste ?
Whatever you decide you should do it explicit. More space or overlapping.
Well! The problem is that I made thin and extra bold for headlines and regular for text, but you say that’s nope, so…
The almost-collision looks a bit ugly
🤔 hm. I will experiment with the spacing some more. Maybe I really have to bend the extra bold back to text usage. Ugh.What’s really annoying is that I recently noticed that the x-height is actually too low for proper text display at UI sizes. I’d have to redo Cantarell. Ugh...
Or maybe continue with what u have. Decide what size would best to use it for. And then make and version with higher x height. Using RMX tools for example to make a quick base.
What's your experience with designing things for text settings? The fonts are invariably going to be used for captions and headings at some point. For example, in gtk#1808 (closed), the main users of Cantarell have tentatively decided on a size range from 9 pt for captions to 24 pt for titles. Is it realistic to support that range without a opsz axis and have the fonts not look like shit near one or the other end of the spectrum?
Typography on the internet is a strange beast. Traditionally you would design a typeface for specific uses. Nowadays due to the availability people tend to use it in many other ways as intended. Often due to ignorance about typography. So the question should actually be, do you want to base your decisions on how the typeface is being used at the moment ?
If you want to create a typeface that will work as intended in every size, I would add an opsz axis. Define up front your size range and work from there. For example 9-24pt. But maybe you might want go even further, micro to display ?
Using a typeface intended for ≈9pt in large size as ≈24pt is doable if you accept that some of the design elements (optical corrections, indents, etc.) will become a stylistic thing.
Current respacing work is done in https://gitlab.gnome.org/GNOME/cantarell-fonts/tree/respacing.